Pilot Testing of SHRP 2 Reliability Data and Analytical Products: Washington (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

116 CHAPTER 8 Pilot Testing and Analysis on SHRP 2 C11 Product 8.1 Introduction Most benefit–cost analysis tools incorporate recurring congestion impacts and exclude nonrecurring (resulting from incident/weather/work zone/demand fluctuation) congestion impacts. This is probably because of the difficulty of estimating nonrecurring congestion impacts. The SHRP 2 program developed the C11 Reliability tool to facilitate estimating both recurring and nonrecurring congestion delays and their associated costs. The tool was applied to analyze the I-5 facility through the Joint Base Lewis-McChord (JBLM), also known as JBLM project. 8.2 Description of the Test Site The research team has selected I-5 facility through the JBLM located between SR 510 (Marvin Road NE) in Lacey and SR 512 in Lakewood for testing the travel time reliability tool. The test site is in Pierce County, Washington State, and is shown in Figure 8.1 (interchange locations are indicated by green circles). This portion of I-5 experiences congestion in both directions of travel particularly during evening peak demand period. A congestion scan from INRIX is shown in Figure 8.2. INRIX data indicates peak period congestion in both directions of travel between Dupont-Steilacoom Road and Thorne Lane. Travel speed drops as low as 35 mph in the northbound direction during part of the evening peak period.

117 Figure 8.1. Map of test site, I-5 through JBLM.

118 Figure 8.2. INRIX congestion scan of I-5 JBLM area. 8.3 Alternatives to Test To test the travel time reliability tool, existing conditions (base case) and six conceptual alternative improvements (scenarios) have been evaluated. These scenarios are: 1. Hard shoulder running between 41st Division Drive and Thorne Lane, and ramp metering with HOV bypass at all interchanges between SR 510 and SR 512; 2. Extend 8 lanes from Berkeley Street interchange to Thorne Lane interchange and provide hard shoulder running between Mounts Road and Berkeley Street interchanges; 3. Add one lane each direction from Mounts Road to Thorne Lane; 4. Hard shoulder running from Mounts Road to Thorne Lane; 5. Ramp metering/increased incident response; and 6. Ramp metering/increased incident response in combination with Option 4. 8.4 Input Data The travel time reliability tool helps perform estimates of travel time and reliability with minimal data. Data entry and scenario management have been made easier by providing a user-friendly

119 interface (Figure 8.3). The tool comes with default data for some of the required data fields while providing options to replace them with project-specific data. Specifically the tool provides default data for the following variables:  Travel time unit costs for personal and commercial travel;  Effect of incident management strategies: o Reduction in incident frequency, and o Reduction in incident duration; and  Reliability ratios (i.e., value of reliability over value of travel time) for personal and commercial travel. Figure 8.3. Data input screen of the travel time reliability tool. The reliability tool allows evaluation of freeway mainline segments between interchanges. For this pilot study, I-5 through JBLM area has been divided into 10 segments (Figure 8.4). Necessary input data for these segments have been collected and/or generated using other tools. Reliability ratios for personal and commercial travel are the default values from the travel time reliability tool.

120 Figure 8.4. Base year (2012) input data. 8.5 Output Data The travel time reliability tool provides different performance metrics in an easy-to-understand format, which aids the users in interpreting and communicating the results of analyses. For example, the tool generates an overall mean TTI, 95th percentile TTI, 80th percentile TTI, 50th percentile TTI, as well as a proportion of trips below 45 and 30 mph speed (an example of output is shown in Figure 8.5). Also performance measures are generated for the base year and a future year assuming an analysis period of 20 years. In addition to performance metrics, the study team developed estimates of travel delay under each conceptual scenario (Figure 8.6). The tool helps estimate congestion delays separately for both personal and commercial travel. All improvement options show reduced congestion delays compared to the base case indicating the tool is sensitive to roadway improvements. “Hard shoulder running with ramp metering and increased incident response” provided most benefits in terms of congestion delay reduction. Begin End NB SB Personal Travel Commercial Travel Reduction in Incident Frequency Reduction in Incident Duration Personal Travel Commercial Travel (mile) (mph) (%) (%) (pcph) (pcph) ($/hour) ($/hour) (%) (%) I-5 Marvin Rd NE (SR 510) to Brown Farm Rd NE 112.01 114.18 2.17 3 3 60 99,000 1.21% 12.06 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Brown Farm Rd NE to Mounts Rd 114.18 116.77 2.59 3 3 60 111,000 1.21% 11.77 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Mounts Rd to Center Dr 116.77 118.02 1.25 3 3 60 120,000 1.21% 11.77 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Center Dr to Dupont-Steilacoom Rd 118.02 119.07 1.05 3 3 60 121,000 0.84% 11.77 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Dupont-Steilacoom Rd to 41st Division Dr 119.07 120.96 1.89 3 3 60 117,000 0.94% 11.77 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 41st Division Dr to Berkeley St 120.96 122.74 1.78 3 3 60 126,000 0.84% 10.08 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Berkeley St to Thorne Lane 122.74 123.64 0.90 3 3 60 134,000 0.94% 10.08 4875 4875 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Thorne Lane to Gravelly Lake Dr. 123.64 124.71 1.07 4 4 60 143,000 0.94% 10.08 6500 6500 Level $22.66 $62.87 2.75 55 0.8 1.1 I-6 Gravelly Lake Dr. to Bridgeport Way 124.71 125.92 1.21 4 4 60 140,000 0.94% 10.08 6500 6500 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Bridgeport Way to SR 512 125.92 127.54 1.62 4 4 60 141,000 0.94% 10.08 6500 6500 Level $22.66 $62.87 2.75 55 0.8 1.1 I-5 Corridor - Marvin Rd to SR 512 112.01 127.54 15.53 Effect of Incident Mgt Reliability Ratio Route Length Posted Speed Annual Traffic Growth Rate Truck Base Year (2012) AADT Travel CostLanesARM Segment Terrain Type SB Capacity NB Capacity

121 S c e n a ri o C o n g e s ti o n M e tr ic s S R 5 1 0 t o B ro w n F a rm R d B ro w n F a rm R d to M o u n ts R d M o u n ts R d t o C e n te r D r C e n te r D r to D u p o n t- S te il a c o o o m R d D u p o n t- S te il a c o o m R d t o 4 1 s t D iv is io n D r 4 1 s t D iv is io n D r to B e rk e le y S t B e rk e le y S t to T h o rn e L a n e T h o rn e L a n e t o G ra v e ll y L a k e D r G ra v e ll y L a k e D r to B ri d g e p o rt W a y B ri d g e p o rt W a y to S R 5 1 2 2 0 1 2 B a s e C a s e Overall mean TTI 1.12 1.10 1.21 1.22 1.16 1.37 1.56 1.18 1.17 1.18 TTI95 1.37 1.34 1.66 1.67 1.51 1.99 2.39 1.53 1.51 1.51 TTI80 1.16 1.13 1.31 1.32 1.23 1.56 1.85 1.27 1.26 1.26 TTI50 1.07 1.05 1.15 1.15 1.10 1.29 1.48 1.14 1.13 1.13 % trips less than 45 mph 13.9% 13.3% 24.0% 24.5% 19.3% 32.6% 41.9% 18.4% 17.6% 17.7% % trips less than 30 mph 2.3% 1.4% 4.5% 4.6% 2.7% 12.8% 23.1% 5.7% 5.5% 5.5% Overall mean TTI 1.07 1.08 1.13 1.13 1.10 1.04 1.06 1.10 1.10 1.10 TTI95 1.25 1.29 1.42 1.43 1.34 1.14 1.21 1.32 1.31 1.31 TTI80 1.10 1.11 1.18 1.18 1.13 1.05 1.08 1.14 1.13 1.13 TTI50 1.04 1.04 1.07 1.08 1.05 1.02 1.03 1.06 1.06 1.06 % trips less than 45 mph 9.7% 11.3% 16.3% 16.5% 13.3% 5.4% 8.1% 12.2% 11.8% 12.0% % trips less than 30 mph 1.2% 1.2% 1.8% 1.8% 1.4% 0.8% 1.0% 1.8% 1.7% 1.7% Overall mean TTI 1.37 1.64 1.87 1.65 1.64 1.77 2.06 1.35 1.33 1.34 TTI95 1.99 2.58 3.06 2.60 2.58 2.86 3.42 1.96 1.92 1.93 TTI80 1.56 1.98 2.33 1.99 1.97 2.18 2.60 1.53 1.51 1.51 TTI50 1.29 1.55 1.77 1.56 1.55 1.68 1.94 1.28 1.26 1.27 % trips less than 45 mph 32.6% 47.6% 58.6% 48.0% 47.6% 54.1% 67.1% 31.9% 30.5% 30.9% % trips less than 30 mph 12.7% 24.8% 33.5% 25.0% 24.8% 29.7% 37.0% 12.9% 12.5% 12.6% Overall mean TTI 1.19 1.58 1.68 1.57 1.56 1.18 1.11 1.22 1.17 1.18 TTI95 1.60 2.44 2.68 2.42 2.39 1.52 1.36 1.64 1.54 1.56 TTI80 1.28 1.89 2.04 1.88 1.85 1.27 1.14 1.32 1.25 1.25 TTI50 1.13 1.49 1.59 1.49 1.48 1.13 1.06 1.16 1.11 1.12 % trips less than 45 mph 22.2% 43.4% 50.1% 43.0% 41.9% 18.2% 13.9% 22.9% 20.3% 20.8% % trips less than 30 mph 3.7% 23.4% 25.9% 23.3% 23.0% 5.6% 1.5% 5.9% 3.0% 3.1% 2 0 1 2 B a s e C a s e 2 0 1 2 S c e n a ri o 1 2 0 3 2 B a s e C a s e 2 0 3 2 S c e n a ri o 1 Figure 8.5. Corridor performance indicators.

122 Figure 8.6. Estimates of annual travel delay. Congestion cost was estimated using the travel time reliability tool. Congestion costs for base case and alternative options are shown in Figure 8.7. The cost of recurring congestion and the cost of unreliability (also known as the cost of nonrecurring congestion) were estimated. The hourly values of travel time for passenger and commercial vehicles were assumed to be $22.66 and $62.87, respectively. The total cost of congestion for 2012 base condition is estimated to be about $31 million (in 2012 dollar values). Figure 8.7. Estimates of annual costs to travelers resulting from congestion.

123 8.6 Cost of Alternatives To perform economic analyses and compare project alternatives, it is necessary to estimate both benefits and costs of alternatives. The travel time reliability tool helps estimate travel time and reliability benefits. Cost estimation of alternatives has been performed using WSDOT’s Planning Level Cost Estimation (PLCE) tool. PLCE is a database tool to perform cost estimation for projects that are very conceptual, often with minimum or no design. The tool has been developed to estimate costs for varieties of projects namely widening existing roadways or bridges, building new roads or bridges, modifying existing interchanges or building new ones, improving intersections, and installing ITSs. PLCE uses a unit price approach that accounts for regional differences as well as differences in land use types and development density within a region. Since unit prices vary by geographic area, separate unit prices are used in the estimate depending on where the project is located. Within each geographic area, unit prices are again a function of density of development such as rural, suburban, urban, and dense urban. The tool comes with default quantities per lane-mile for common items such as grading, drainage, pavement, traffic control, etc. The underlying assumption of the methodology is that little or no geotechnical data are known at the time of planning-level estimate. Furthermore, the tool comes with default unit costs obtained from historical data of WSDOT’s past projects. Some unit prices were adjusted for differences in area prices, terrain (e.g., level, rolling, or mountainous), ground conditions, and design assumptions. These unit costs can be easily edited through user-friendly interfaces. An example of selecting project components to be included in the estimation is shown in Figure 8.8. (Additional information about the tool is available at http://www.wsdot.wa.gov/mapsdata/travel/pdf/PLCEManual_12- 12-2012.pdf.) A summary of estimated costs of alternatives is presented in Figure 8.9. Ramp metering and incident response (Scenario 5) would cost the least, while adding a lane in each direction between Mounts Road and Thorne Lane (Scenario 3) would cost the most. Scenario 3 requires the addition of two new lanes and reconstruction of a few interchanges and bridges, resulting in a much higher cost compared to other scenarios.

124 Figure 8.8. Main menu of the PLCE tool. Figure 8.9. Estimated costs of alternatives. 8.7 Benefit–Cost Analysis The travel time reliability tool performs an estimation of travel benefits. However, it does not facilitate performing benefit–cost analysis incorporating project costs and benefits. This analysis has been conducted outside the reliability tool using methodology in WSDOT’s benefit–cost analysis tool (known as MP3B-C tool). This tool was found to be suitable for conducting benefit–cost analysis for the type of projects being analyzed and available data.

125 A summary of the benefit–cost analysis is shown in Figure 8.10. The analysis was performed with a set of assumptions that include the following:  An analysis period of 20 years;  Annual discount rate of 4% (used to convert future costs and benefits to present values);  Benefits include travel time savings and reduction of unreliability;  Personal and commercial travel time values are $22.66 and $62.87 per hour, respectively;  Residual values were used to adjust the benefit–cost ratio to account for the value of the improvement remaining after 20 years (the residual value methodology is based on work done for AASHTO by the Texas Transportation Institute) and was done by applying the following factors to the project’s estimated costs: o Right of way, 0.45, o Grading and drainage, 0.40, o Structures, 0.43, and o All other costs (including PE), 0.00;  Annual roadway operations and maintenance cost is $16,500 (in 2012 dollar values) per lane-mile;  Annual IRT cost is $7,000 (in 2012 dollar values) per lane-mile; and  Annual signal/ramp meter operations and maintenance cost is $1,200 (in 2012 dollar values).

126 Figure 8.10. Summary of benefit–cost analysis. To prepare TIGER III Grant Application for I-5 JBLM project, WSDOT conducted an economic analysis using TREDIS software. The total project benefit–cost ratio, based on anticipated project design and construction costs, as well as on all monetized benefits, including travel time, vehicle operating costs, reliability, safety, freight, and environmental, was estimated to range from 5.67 to 8.38. Travel time and reliability benefits were estimated to amount to $123.8 million (undiscounted) for 24 years. The travel time reliability tool provides estimates of benefits that include recurring congestion reduction and reliability improvements. When analyzed the same JBLM project using the travel time reliability tool with roadway capacity (2,190 pcphpl) from the HCM (as suggested by the tool), the benefit–cost ratio ranged from 1.96 to 2.43. Given this tool is considering only direct benefits from travel time and reliability improvements, the values are expected to be somewhat lower than those from analyses for TIGER III Grant Application (using TREDIS software). However, the benefit–cost ratios from the travel time reliability tool seem to be too low when compared with the values from TIGER III Grant Application. When researchers analyzed the same JBLM project using the travel time reliability tool with reduced roadway capacity (1,625 pcphpl), the benefit–cost ratio ranged from 22.23 to 27.56. In this case, the benefit–-cost ratios from the travel time reliability tool are found to be much higher than the values from TIGER III Grant Application. 8.8 Validation of Outputs from the Travel Time Reliability Tool Validation of outputs from the reliability tool was done by comparing the base year outputs, particularly total travel delay and delay cost, from this tool to the similar data from INRIX

127 analytic tools (Figure 8.11). (More information about the INRIX Traffic Analytic Tools is available at http://www.itproportal.com/2013/09/21/a-closer-look-at-inrix-the-worlds-largest- traffic-intelligence-network/#ixzz2hubtfXAB.) Figure 8.11. Snapshot of INRIX Traffic Analytic Tools. INRIX recently added a new module called “User Delay Cost Analysis” to generate travel delay costs for each hour of a day for 365 days. For maintaining consistency of data, cost of congestion was estimated using INRIX analytic tools by applying the same hourly value of travel time for passenger and commercial vehicles as were assumed in the travel time reliability tool. The 2012 annual weekday cost of congestion from INRIX was $17,192,000 (in 2012 dollar values), while the travel time reliability tool showed a value of $1,720,000 when HCM capacity was used and a value of $31,135,000 when reduced capacity (1,625 pcphpl) was used. While using HCM capacity in the travel time reliability tool, INRIX data indicated about 10 times higher congestion cost than that from the travel time reliability tool. In contrast, INRIX data indicated about 45% lower congestion cost than that from the travel time reliability tool with reduced capacity. For validation purposes TTI data from both the travel time reliability tool and INRIX were compared. The reliability tool with HCM capacity indicates less severe congestion than indicated by INRIX data. An example of TTI values between Berkeley Street and Thorne Lane is shown in Figure 8.12. It is also observed that TTI values from the reliability tool are more or less the same (close to 1 indicating not much of congestion) during both peak and off-peak periods.

128 Note that the reliability tool provides an overall TTI for both direction of travel instead of providing separate indices for each direction. Figure 8.12. TTI values for 2012 base case using HCM capacity. To further investigate if the tool underestimates congestion or it is because of inaccurate data entered into the tool, researchers rechecked the data used in the first round of analyses. No data issues were found. Then additional tests were conducted on I-405 between I-90 and 8th Street SE. These additional tests also indicated lower than expected congestion (i.e., TTI values). In addition sensitivity analyses were performed by inputting lower capacity than that calculated using HCM methodologies. When reduced roadway capacity (e.g., congested capacity) is used, the reliability tool produces higher TTI values and indicates sensitivity to time of day. For example, a comparison of 2012 TTI values from INRIX and the reliability tool is presented in Figure 8.13 for the same I-5 segment between Berkeley Street and Thorne Lane.

129 Figure 8.13. TTI values for 2012 base case using congested capacity. If TTI values are generated by direction as well as by time of day, it becomes easier to understand which direction of travel experiences congestion effects at what time of the day. For example, INRIX data indicate relatively higher congestion in northbound direction during p.m. peak period (3:00 p.m. to 7:00 p.m.). The reliability tool does not show TTI values by direction, and therefore it is not possible to assess which direction of travel experiences what level of congestion at what time of the day. 8.9 Assessment of the Travel Time Reliability Tool The research team conducted an assessment of the travel time reliability tool for its input requirements, ease of use, calculation algorithms, usefulness and organization of output data, scenario management, and reasonableness of the results produced by the tool. A summary of the assessment is provided below. 8.10 General Observations The travel time reliability tool requires minimal data and appears to be easy to use. The tool has been designed to require data that can be easily collected or assembled by those conducting a sketch planning study. The required data can be acquired from widely used data sources. The tool comes with simple and easy scenario management features. The tool facilitates analyses of multiple scenarios by allowing creating and saving new scenarios with relative ease. The tool displays results of the base and alternative scenarios side by side for ease of comparison. This tool allows users to perform quick assessment of the effects of highway investments. It allows conducting assessment of transportation investment benefits in terms of reducing

130 recurring delay as well as improving travel time reliability. Most of the existing economic analysis tools consider only recurring delay while excluding the effects of travel time reliability. Since this tool accounts for this additional benefit from travel time reliability, it is expected to show more positive effects of a highway investment on the economy than typical estimates using traditional tools and methodologies. The tool was tested on a wide range of improvement options. A few observations regarding the analysis results are:  The tool estimates travel delay that is about one-tenth of the values from INRIX traffic analysis tools. It seems like the tool underestimates travel impacts. This could be because of the fact that the tool does not account for impacts from traffic volume other than mainline volume, although ramp spacing and ramp traffic volume may have considerable effect on freeway operations. Particularly the I-5 ramp traffic volume along JBLM is thought to be the primary cause of congested condition along this stretch of the facility, but the tool does not analyze the freeway mainline and ramps together as a system.  The travel time reliability tool uses three sets of hourly traffic distribution factors for peak travel direction of a roadway. The tool selects one of these three sets based on AADT/capacity ratio: less than 7.0, 7.1 to 11.0, and greater than 11.0. Base case and an improvement option could sometimes have different AADT/capacity ratio leading to usage of a different set of hourly distribution factors, and thus an improvement option might sometimes show worse traffic congestion than the base case. For example, the study included a 6-lane freeway segment with AADT of 111,000. The roadway capacity (in this case researchers used congested capacity) for the base case was 9,750 pcph (passenger cars per hour) and that of the improvement option was 10,285 pcph (assuming 5.5% increase of capacity because of ramp meters and HOV bypass lanes). This combination of AADT and capacity generates AADT/capacity ratios of 11.38 and 10.79 for the base case and the alternative. These ratios lead to use of different hourly distribution sets for the base case and alternative option resulting in higher TTI values for the alternative option (overall mean TTI values of 1.10 for the base case and 1.22 for the alternative option) even though the alternative option has higher capacity and expected to reduce congestion. In this case, the tool indicates congestion would increase even though traffic carrying capacity of the freeway is being increased.  When roadway capacity based on HCM was used (as suggested by the tool), the nonrecurring congestion delay appeared to be much higher than that of recurring congestion for all improvement scenarios. However, when reduced roadway capacity was used, the tool produced nonrecurring congestion delay ranging from 8% to 19% for the scenarios, which is more in line with the expectation. Note that a 2003 report by Washington Transportation Center titled Measurement of Recurring versus Nonrecurring Congestion: Technical Report shows nonrecurring congestion ranging from 5% to 58%

131 depending on type of estimate (e.g., conservative or liberal). This report is found at http://www.wsdot.wa.gov/research/reports/fullreports/568.1.pdf. 8.11 Applicability In assessing the tool, special attention was given to the applicability of the tool to evaluate various improvement scenarios. An overview of the assessment follows:  The travel time reliability tool requires minimal data for performing assessment of impacts of highway investments. Most of the data the tool requires seem to be relatively easy to gather. So the tool can easily be used as a sketch planning tool for analysis of travel time and reliability effects of some of the conceptual improvements typically analyzed as part of planning studies.  In assessing travel benefits, the travel time reliability tool accounts for impacts of reduced incident frequency and duration resulting from incident management strategies. However, it does not provide any default input values or any sources or references to get help in developing input data. The effects of incident management strategies have to be estimated outside this tool and then entered as input into this tool.  The calculation methodology is directly applicable only to a roadway mainline (segments between interchanges/intersections), not to improvements at roadway intersections, interchanges, and freeway ramps. Therefore, it may not provide a comprehensive assessment of transportation options, because it does not perform analysis on a system of freeway mainline, ramps, and connecting roads accounting for vehicle interactions at the junctions.  The tool has been designed to evaluate roadway capacity improvements (e.g., adding lanes). It does not come with a methodology to estimate benefits from varieties of transportation improvement types including ITS improvements, demand management strategies, etc. Therefore, this tool does not seem to be applicable to analysis of all sorts of transportation improvements typically considered by an agency.  This tool does not perform any benefit–cost analysis; it just produces travel time and reliability benefits that can be used in a benefit–cost analysis. So for comparing alternatives, further economic analyses need to be performed using other appropriate tools.